Personalized recommendation method and system for friends and applications

Abstract

The invention provides a personalized recommendation method and system for friends and applications. The method comprises the steps that A, a recommendation score S is calculated; B, recommended content is displayed according to the recommendation score; C, a recommendation weight vector is adjusted according to feedback information about the recommended content from a user; D, the recommended content is updated according to the adjusted user recommendation weight vector. The personalized recommendation method and system have the advantages that the application range is wider, recommendation results are more accurate, the calculating speed is higher, calculation at idle time is achieved, a blacklist mechanism is adopted, the accuracy of information recommendation can be effectively improved, calculation cost is saved, and a good recommendation effect is achieved.

Description

technical field [0001] The invention relates to the technical field of network communication, in particular to a Hadoop-based friend and application recommendation method and system in a social network. Background technique [0002] Social networks such as Renren, Facebook, and Friends have become an indispensable way of socializing. The number of active users is very important for social networking sites, and helping users establish and expand a stable relationship network is the most effective way to maintain and increase the number of active users. This requires recommending potential friends and applications that can interact with friends for users. [0003] However, most of the existing methods for recommending friends and applications are single. Current friend recommendation methods can be divided into several categories: 1. According to the number of common friends among non-friend users. 2. According to the similarity of personal data (graduation school, place of...

AI Technical Summary

Problems solved by technology

2) Does not consider the interactive information between users

Interaction is the basic attribute of social network. Without considering the interaction between users, it is impossible to get accurate and reasonable recommendation results.

3) It is easy to cause user information leakage

This method of application recommendation based on user history behavior has the following disadvantages: 1) When the user has not added any application, this method will be useless and there will be no recommendation results

2) It will generate more recommendation results that are duplicated with the user's added application functions

3) The application usage of friends is not considered, and accurate and reasonable recommendation results cannot be obtained

[0010] 1. Friend recommendation considers latitude: This method essentially recommends friends based on the latitude of the number of mutual friends. This method has several disadvantages: 1. It is too one-sided, focusing on the number of mutual friends, and does not distinguish between friends and users. It does not match the fact that "users who have a high degree of mutual friends are more likely to become friends than users who have more mutual friends"

2. For new registration, users who have filled in personal information but have not added friends, this method cannot be used

[0011] 2. Blacklist and scrolling display mechanism: The above method does not use the blacklist mechanism, which leads to recommended friends who rank high but users do not want to add occupy the recommendation section for a long time, making the recommendation mechanism meaningless

[0012] 3. Whether the user needs to take the initiative to intervene: the above method requires the user to actively set the group and set the group level. For users who are not used to grouping friends or the grouping is inaccurate, this method is very ineffective or even unusable

[0017] 1. Friend recommendation considers latitude: This patent document uses a single latitude for friend recommendation, which is too single to consider, and uses the user’s check-in information, which will have the following problems: (1) The method is suitable for limited situations: only a few social The website provides a sign-in function

(2) Leakage of user information: The check-in information usually includes the user's location information, so the similarity between the user's check-in location information displayed in the top user and the user's check-in location information may be very high, resulting in information leakage

[0024] 2. Applicable scenarios: The applicable scenarios of this method are limited. When the user has not added any applications, this method will lose its effect

[0025] 3. Blacklist mechanism: This method will still cause the top-ranked apps that users do not want to add to always occupy the app recommendation display module, while the apps added by users with lower ranks are more likely to be unable to be displayed, resulting in meaningless app recommendations

However, this invention is only recommended based on the information of the current user, which does not conform to the social characteristics

Images